Do I use AI in my work as an English–Polish translator? Yes — for research, exploring ideas and comparing ways of expressing a message. But the useful question is not simply whether I use it. It is how I use it, and where I draw the line.
From typewriters to the Tube
When I was studying towards an MA in Translation at the University of Westminster in 2005–2006, my lecturers sometimes described their earlier working lives: typing translations on a typewriter and, when speed mattered, jumping on the Tube to deliver the finished copy in person.
It is a story that has stayed with me. For all our discussions about translation technology, sometimes the fastest delivery system was the London Underground.
Today, asking a translator whether they use AI feels to me a little like asking ten years ago whether they used Google, or twenty years ago whether they used a computer. Technology belongs in our professional toolkit. That does not make every use appropriate, or remove a client’s right to ask questions.
What I actually use AI for
Research is one example. AI can help me explore a subject, identify terminology to investigate and frame better questions. I treat its suggestions as starting points, not evidence: a convincing answer still needs checking against reliable sources.
It is also a useful sounding board. I can explore several approaches to a headline, consider a different tone or test an idea against a particular audience. I decide which suggestions work in Polish, which miss the point and which should be discarded.
I also use it to develop and compare alternative wording, including options for A/B testing. There is an important distinction here: asking AI to compare two versions is not the same as testing them with real readers. Claims about which version performs better need audience data, not just a model’s preference.
Used this way, AI helps me explore more possibilities and spend less time on preliminary work. It does not make the final linguistic decision for me.
Efficiency without handing over responsibility
Some people imagine that using AI means handing the translation over to the machine. In practice, I use it as one more professional tool: something that can make parts of the work faster and more efficient without replacing the judgement behind the final text. Meaning, context, tone and suitability for the reader remain my responsibility. A fluent suggestion is not automatically an accurate or appropriate one.
Confidentiality is part of that responsibility. I do not upload client documents or data to AI tools without the client’s consent. Permission is not a substitute for checking the tool, the project requirements and any confidentiality restrictions. Researching a general topic is not the same as sharing the document I have been asked to translate.
Translators should help shape how technology is used
I believe translators should be at the forefront of technology in our profession: trying tools, understanding their limits and helping clients use them sensibly. That is different from adopting every new feature uncritically.
We do not serve clients better by clinging to a typewriter. Nor do we serve them better by accepting everything a machine produces.
The aim is to use better tools while keeping the judgement, care and accountability that make the work professional.
For a practical summary of my approach and client-consent policy, see “Do you use AI in your work?” in my Translation Process FAQ.